feat: Centralized Request Processing middleware (#3847)

* squash past, centralize request middleware PR

Signed-off-by: Dave Lee <dave@gray101.com>

* migrate bruno request files to examples repo

Signed-off-by: Dave Lee <dave@gray101.com>

* fix

Signed-off-by: Dave Lee <dave@gray101.com>

* Update tests/e2e-aio/e2e_test.go

Signed-off-by: Ettore Di Giacinto <mudler@users.noreply.github.com>

---------

Signed-off-by: Dave Lee <dave@gray101.com>
Signed-off-by: Ettore Di Giacinto <mudler@users.noreply.github.com>
Co-authored-by: Ettore Di Giacinto <mudler@users.noreply.github.com>

core/http/endpoints/openai/chat.go

@@ -5,18 +5,19 @@ import (
 	"bytes"
 	"encoding/json"
 	"fmt"
-	"strings"
 	"time"
 
 	"github.com/gofiber/fiber/v2"
 	"github.com/google/uuid"
 	"github.com/mudler/LocalAI/core/backend"
 	"github.com/mudler/LocalAI/core/config"
+	"github.com/mudler/LocalAI/core/http/middleware"
 	"github.com/mudler/LocalAI/core/schema"
 	"github.com/mudler/LocalAI/pkg/functions"
+
+	"github.com/mudler/LocalAI/pkg/model"
 	"github.com/mudler/LocalAI/pkg/templates"
 
-	model "github.com/mudler/LocalAI/pkg/model"
 	"github.com/rs/zerolog/log"
 	"github.com/valyala/fasthttp"
 )
@@ -174,26 +175,20 @@ func ChatEndpoint(cl *config.BackendConfigLoader, ml *model.ModelLoader, evaluat
 		textContentToReturn = ""
 		id = uuid.New().String()
 		created = int(time.Now().Unix())
-		// Set CorrelationID
-		correlationID := c.Get("X-Correlation-ID")
-		if len(strings.TrimSpace(correlationID)) == 0 {
-			correlationID = id
+		
+		input, ok := c.Locals(middleware.CONTEXT_LOCALS_KEY_LOCALAI_REQUEST).(*schema.OpenAIRequest)
+		if !ok || input.Model == "" {
+			return fiber.ErrBadRequest
 		}
-		c.Set("X-Correlation-ID", correlationID)
 
-		// Opt-in extra usage flag
 		extraUsage := c.Get("Extra-Usage", "") != ""
 
-		modelFile, input, err := readRequest(c, cl, ml, startupOptions, true)
-		if err != nil {
-			return fmt.Errorf("failed reading parameters from request:%w", err)
+		config, ok := c.Locals(middleware.CONTEXT_LOCALS_KEY_MODEL_CONFIG).(*config.BackendConfig)
+		if !ok || config == nil {
+			return fiber.ErrBadRequest
 		}
 
-		config, input, err := mergeRequestWithConfig(modelFile, input, cl, ml, startupOptions.Debug, startupOptions.Threads, startupOptions.ContextSize, startupOptions.F16)
-		if err != nil {
-			return fmt.Errorf("failed reading parameters from request:%w", err)
-		}
-		log.Debug().Msgf("Configuration read: %+v", config)
+		log.Debug().Msgf("Chat endpoint configuration read: %+v", config)
 
 		funcs := input.Functions
 		shouldUseFn := len(input.Functions) > 0 && config.ShouldUseFunctions()
@@ -543,7 +538,7 @@ func handleQuestion(config *config.BackendConfig, input *schema.OpenAIRequest, m
 		audios = append(audios, m.StringAudios...)
 	}
 
-	predFunc, err := backend.ModelInference(input.Context, prompt, input.Messages, images, videos, audios, ml, *config, o, nil)
+	predFunc, err := backend.ModelInference(input.Context, prompt, input.Messages, images, videos, audios, ml, config, o, nil)
 	if err != nil {
 		log.Error().Err(err).Msg("model inference failed")
 		return "", err

core/http/endpoints/openai/completion.go

@@ -10,12 +10,13 @@ import (
 
 	"github.com/mudler/LocalAI/core/backend"
 	"github.com/mudler/LocalAI/core/config"
+	"github.com/mudler/LocalAI/core/http/middleware"
 
 	"github.com/gofiber/fiber/v2"
 	"github.com/google/uuid"
 	"github.com/mudler/LocalAI/core/schema"
 	"github.com/mudler/LocalAI/pkg/functions"
-	model "github.com/mudler/LocalAI/pkg/model"
+	"github.com/mudler/LocalAI/pkg/model"
 	"github.com/mudler/LocalAI/pkg/templates"
 	"github.com/rs/zerolog/log"
 	"github.com/valyala/fasthttp"
@@ -26,11 +27,11 @@ import (
 // @Param request body schema.OpenAIRequest true "query params"
 // @Success 200 {object} schema.OpenAIResponse "Response"
 // @Router /v1/completions [post]
+
 func CompletionEndpoint(cl *config.BackendConfigLoader, ml *model.ModelLoader, evaluator *templates.Evaluator, appConfig *config.ApplicationConfig) func(c *fiber.Ctx) error {
-	id := uuid.New().String()
 	created := int(time.Now().Unix())
 
-	process := func(s string, req *schema.OpenAIRequest, config *config.BackendConfig, loader *model.ModelLoader, responses chan schema.OpenAIResponse, extraUsage bool) {
+	process := func(id string, s string, req *schema.OpenAIRequest, config *config.BackendConfig, loader *model.ModelLoader, responses chan schema.OpenAIResponse, extraUsage bool) {
 		ComputeChoices(req, s, config, appConfig, loader, func(s string, c *[]schema.Choice) {}, func(s string, tokenUsage backend.TokenUsage) bool {
 			usage := schema.OpenAIUsage{
 				PromptTokens:     tokenUsage.Prompt,
@@ -63,22 +64,18 @@ func CompletionEndpoint(cl *config.BackendConfigLoader, ml *model.ModelLoader, e
 	}
 
 	return func(c *fiber.Ctx) error {
-		// Add Correlation
-		c.Set("X-Correlation-ID", id)
-
-		// Opt-in extra usage flag
+		// Handle Correlation
+		id := c.Get("X-Correlation-ID", uuid.New().String())
 		extraUsage := c.Get("Extra-Usage", "") != ""
 
-		modelFile, input, err := readRequest(c, cl, ml, appConfig, true)
-		if err != nil {
-			return fmt.Errorf("failed reading parameters from request:%w", err)
+		input, ok := c.Locals(middleware.CONTEXT_LOCALS_KEY_LOCALAI_REQUEST).(*schema.OpenAIRequest)
+		if !ok || input.Model == "" {
+			return fiber.ErrBadRequest
 		}
 
-		log.Debug().Msgf("`input`: %+v", input)
-
-		config, input, err := mergeRequestWithConfig(modelFile, input, cl, ml, appConfig.Debug, appConfig.Threads, appConfig.ContextSize, appConfig.F16)
-		if err != nil {
-			return fmt.Errorf("failed reading parameters from request:%w", err)
+		config, ok := c.Locals(middleware.CONTEXT_LOCALS_KEY_MODEL_CONFIG).(*config.BackendConfig)
+		if !ok || config == nil {
+			return fiber.ErrBadRequest
 		}
 
 		if config.ResponseFormatMap != nil {
@@ -122,7 +119,7 @@ func CompletionEndpoint(cl *config.BackendConfigLoader, ml *model.ModelLoader, e
 
 			responses := make(chan schema.OpenAIResponse)
 
-			go process(predInput, input, config, ml, responses, extraUsage)
+			go process(id, predInput, input, config, ml, responses, extraUsage)
 
 			c.Context().SetBodyStreamWriter(fasthttp.StreamWriter(func(w *bufio.Writer) {
 

core/http/endpoints/openai/edit.go

@@ -2,16 +2,17 @@ package openai
 
 import (
 	"encoding/json"
-	"fmt"
 	"time"
 
 	"github.com/mudler/LocalAI/core/backend"
 	"github.com/mudler/LocalAI/core/config"
+	"github.com/mudler/LocalAI/core/http/middleware"
 
 	"github.com/gofiber/fiber/v2"
 	"github.com/google/uuid"
 	"github.com/mudler/LocalAI/core/schema"
-	model "github.com/mudler/LocalAI/pkg/model"
+
+	"github.com/mudler/LocalAI/pkg/model"
 	"github.com/mudler/LocalAI/pkg/templates"
 
 	"github.com/rs/zerolog/log"
@@ -25,20 +26,21 @@ import (
 func EditEndpoint(cl *config.BackendConfigLoader, ml *model.ModelLoader, evaluator *templates.Evaluator, appConfig *config.ApplicationConfig) func(c *fiber.Ctx) error {
 
 	return func(c *fiber.Ctx) error {
+
+		input, ok := c.Locals(middleware.CONTEXT_LOCALS_KEY_LOCALAI_REQUEST).(*schema.OpenAIRequest)
+		if !ok || input.Model == "" {
+			return fiber.ErrBadRequest
+		}
 		// Opt-in extra usage flag
 		extraUsage := c.Get("Extra-Usage", "") != ""
 
-		modelFile, input, err := readRequest(c, cl, ml, appConfig, true)
-		if err != nil {
-			return fmt.Errorf("failed reading parameters from request:%w", err)
+		config, ok := c.Locals(middleware.CONTEXT_LOCALS_KEY_MODEL_CONFIG).(*config.BackendConfig)
+		if !ok || config == nil {
+			return fiber.ErrBadRequest
 		}
 
-		config, input, err := mergeRequestWithConfig(modelFile, input, cl, ml, appConfig.Debug, appConfig.Threads, appConfig.ContextSize, appConfig.F16)
-		if err != nil {
-			return fmt.Errorf("failed reading parameters from request:%w", err)
-		}
-
-		log.Debug().Msgf("Parameter Config: %+v", config)
+		log.Debug().Msgf("Edit Endpoint Input : %+v", input)
+		log.Debug().Msgf("Edit Endpoint Config: %+v", *config)
 
 		var result []schema.Choice
 		totalTokenUsage := backend.TokenUsage{}

core/http/endpoints/openai/embeddings.go

@@ -2,11 +2,11 @@ package openai
 
 import (
 	"encoding/json"
-	"fmt"
 	"time"
 
 	"github.com/mudler/LocalAI/core/backend"
 	"github.com/mudler/LocalAI/core/config"
+	"github.com/mudler/LocalAI/core/http/middleware"
 	"github.com/mudler/LocalAI/pkg/model"
 
 	"github.com/google/uuid"
@@ -23,14 +23,14 @@ import (
 // @Router /v1/embeddings [post]
 func EmbeddingsEndpoint(cl *config.BackendConfigLoader, ml *model.ModelLoader, appConfig *config.ApplicationConfig) func(c *fiber.Ctx) error {
 	return func(c *fiber.Ctx) error {
-		model, input, err := readRequest(c, cl, ml, appConfig, true)
-		if err != nil {
-			return fmt.Errorf("failed reading parameters from request:%w", err)
+		input, ok := c.Locals(middleware.CONTEXT_LOCALS_KEY_LOCALAI_REQUEST).(*schema.OpenAIRequest)
+		if !ok || input.Model == "" {
+			return fiber.ErrBadRequest
 		}
 
-		config, input, err := mergeRequestWithConfig(model, input, cl, ml, appConfig.Debug, appConfig.Threads, appConfig.ContextSize, appConfig.F16)
-		if err != nil {
-			return fmt.Errorf("failed reading parameters from request:%w", err)
+		config, ok := c.Locals(middleware.CONTEXT_LOCALS_KEY_MODEL_CONFIG).(*config.BackendConfig)
+		if !ok || config == nil {
+			return fiber.ErrBadRequest
 		}
 
 		log.Debug().Msgf("Parameter Config: %+v", config)

core/http/endpoints/openai/image.go

@@ -15,6 +15,7 @@ import (
 
 	"github.com/google/uuid"
 	"github.com/mudler/LocalAI/core/config"
+	"github.com/mudler/LocalAI/core/http/middleware"
 	"github.com/mudler/LocalAI/core/schema"
 
 	"github.com/mudler/LocalAI/core/backend"
@@ -66,25 +67,23 @@ func downloadFile(url string) (string, error) {
 // @Router /v1/images/generations [post]
 func ImageEndpoint(cl *config.BackendConfigLoader, ml *model.ModelLoader, appConfig *config.ApplicationConfig) func(c *fiber.Ctx) error {
 	return func(c *fiber.Ctx) error {
-		m, input, err := readRequest(c, cl, ml, appConfig, false)
-		if err != nil {
-			return fmt.Errorf("failed reading parameters from request:%w", err)
+		input, ok := c.Locals(middleware.CONTEXT_LOCALS_KEY_LOCALAI_REQUEST).(*schema.OpenAIRequest)
+		if !ok || input.Model == "" {
+			log.Error().Msg("Image Endpoint - Invalid Input")
+			return fiber.ErrBadRequest
 		}
-
-		if m == "" {
-			m = "stablediffusion"
-		}
-		log.Debug().Msgf("Loading model: %+v", m)
-
-		config, input, err := mergeRequestWithConfig(m, input, cl, ml, appConfig.Debug, 0, 0, false)
-		if err != nil {
-			return fmt.Errorf("failed reading parameters from request:%w", err)
+		
+		config, ok := c.Locals(middleware.CONTEXT_LOCALS_KEY_MODEL_CONFIG).(*config.BackendConfig)
+		if !ok || config == nil {
+			log.Error().Msg("Image Endpoint - Invalid Config")
+			return fiber.ErrBadRequest
 		}
 
 		src := ""
 		if input.File != "" {
 
 			fileData := []byte{}
+			var err error
 			// check if input.File is an URL, if so download it and save it
 			// to a temporary file
 			if strings.HasPrefix(input.File, "http://") || strings.HasPrefix(input.File, "https://") {

core/http/endpoints/openai/inference.go

@@ -37,7 +37,7 @@ func ComputeChoices(
 	}
 
 	// get the model function to call for the result
-	predFunc, err := backend.ModelInference(req.Context, predInput, req.Messages, images, videos, audios, loader, *config, o, tokenCallback)
+	predFunc, err := backend.ModelInference(req.Context, predInput, req.Messages, images, videos, audios, loader, config, o, tokenCallback)
 	if err != nil {
 		return result, backend.TokenUsage{}, err
 	}

core/http/endpoints/openai/request.go

@@ -1,326 +0,0 @@
-package openai
-
-import (
-	"context"
-	"encoding/json"
-	"fmt"
-	"strconv"
-
-	"github.com/gofiber/fiber/v2"
-	"github.com/google/uuid"
-	"github.com/mudler/LocalAI/core/config"
-	fiberContext "github.com/mudler/LocalAI/core/http/ctx"
-	"github.com/mudler/LocalAI/core/schema"
-	"github.com/mudler/LocalAI/pkg/functions"
-	"github.com/mudler/LocalAI/pkg/model"
-	"github.com/mudler/LocalAI/pkg/templates"
-	"github.com/mudler/LocalAI/pkg/utils"
-	"github.com/rs/zerolog/log"
-)
-
-type correlationIDKeyType string
-
-// CorrelationIDKey to track request across process boundary
-const CorrelationIDKey correlationIDKeyType = "correlationID"
-
-func readRequest(c *fiber.Ctx, cl *config.BackendConfigLoader, ml *model.ModelLoader, o *config.ApplicationConfig, firstModel bool) (string, *schema.OpenAIRequest, error) {
-	input := new(schema.OpenAIRequest)
-
-	// Get input data from the request body
-	if err := c.BodyParser(input); err != nil {
-		return "", nil, fmt.Errorf("failed parsing request body: %w", err)
-	}
-
-	received, _ := json.Marshal(input)
-	// Extract or generate the correlation ID
-	correlationID := c.Get("X-Correlation-ID", uuid.New().String())
-
-	ctx, cancel := context.WithCancel(o.Context)
-	// Add the correlation ID to the new context
-	ctxWithCorrelationID := context.WithValue(ctx, CorrelationIDKey, correlationID)
-
-	input.Context = ctxWithCorrelationID
-	input.Cancel = cancel
-
-	log.Debug().Msgf("Request received: %s", string(received))
-
-	modelFile, err := fiberContext.ModelFromContext(c, cl, ml, input.Model, firstModel)
-
-	return modelFile, input, err
-}
-
-func updateRequestConfig(config *config.BackendConfig, input *schema.OpenAIRequest) {
-	if input.Echo {
-		config.Echo = input.Echo
-	}
-	if input.TopK != nil {
-		config.TopK = input.TopK
-	}
-	if input.TopP != nil {
-		config.TopP = input.TopP
-	}
-
-	if input.Backend != "" {
-		config.Backend = input.Backend
-	}
-
-	if input.ClipSkip != 0 {
-		config.Diffusers.ClipSkip = input.ClipSkip
-	}
-
-	if input.ModelBaseName != "" {
-		config.AutoGPTQ.ModelBaseName = input.ModelBaseName
-	}
-
-	if input.NegativePromptScale != 0 {
-		config.NegativePromptScale = input.NegativePromptScale
-	}
-
-	if input.UseFastTokenizer {
-		config.UseFastTokenizer = input.UseFastTokenizer
-	}
-
-	if input.NegativePrompt != "" {
-		config.NegativePrompt = input.NegativePrompt
-	}
-
-	if input.RopeFreqBase != 0 {
-		config.RopeFreqBase = input.RopeFreqBase
-	}
-
-	if input.RopeFreqScale != 0 {
-		config.RopeFreqScale = input.RopeFreqScale
-	}
-
-	if input.Grammar != "" {
-		config.Grammar = input.Grammar
-	}
-
-	if input.Temperature != nil {
-		config.Temperature = input.Temperature
-	}
-
-	if input.Maxtokens != nil {
-		config.Maxtokens = input.Maxtokens
-	}
-
-	if input.ResponseFormat != nil {
-		switch responseFormat := input.ResponseFormat.(type) {
-		case string:
-			config.ResponseFormat = responseFormat
-		case map[string]interface{}:
-			config.ResponseFormatMap = responseFormat
-		}
-	}
-
-	switch stop := input.Stop.(type) {
-	case string:
-		if stop != "" {
-			config.StopWords = append(config.StopWords, stop)
-		}
-	case []interface{}:
-		for _, pp := range stop {
-			if s, ok := pp.(string); ok {
-				config.StopWords = append(config.StopWords, s)
-			}
-		}
-	}
-
-	if len(input.Tools) > 0 {
-		for _, tool := range input.Tools {
-			input.Functions = append(input.Functions, tool.Function)
-		}
-	}
-
-	if input.ToolsChoice != nil {
-		var toolChoice functions.Tool
-
-		switch content := input.ToolsChoice.(type) {
-		case string:
-			_ = json.Unmarshal([]byte(content), &toolChoice)
-		case map[string]interface{}:
-			dat, _ := json.Marshal(content)
-			_ = json.Unmarshal(dat, &toolChoice)
-		}
-		input.FunctionCall = map[string]interface{}{
-			"name": toolChoice.Function.Name,
-		}
-	}
-
-	// Decode each request's message content
-	imgIndex, vidIndex, audioIndex := 0, 0, 0
-	for i, m := range input.Messages {
-		nrOfImgsInMessage := 0
-		nrOfVideosInMessage := 0
-		nrOfAudiosInMessage := 0
-
-		switch content := m.Content.(type) {
-		case string:
-			input.Messages[i].StringContent = content
-		case []interface{}:
-			dat, _ := json.Marshal(content)
-			c := []schema.Content{}
-			json.Unmarshal(dat, &c)
-
-			textContent := ""
-			// we will template this at the end
-
-		CONTENT:
-			for _, pp := range c {
-				switch pp.Type {
-				case "text":
-					textContent += pp.Text
-					//input.Messages[i].StringContent = pp.Text
-				case "video", "video_url":
-					// Decode content as base64 either if it's an URL or base64 text
-					base64, err := utils.GetContentURIAsBase64(pp.VideoURL.URL)
-					if err != nil {
-						log.Error().Msgf("Failed encoding video: %s", err)
-						continue CONTENT
-					}
-					input.Messages[i].StringVideos = append(input.Messages[i].StringVideos, base64) // TODO: make sure that we only return base64 stuff
-					vidIndex++
-					nrOfVideosInMessage++
-				case "audio_url", "audio":
-					// Decode content as base64 either if it's an URL or base64 text
-					base64, err := utils.GetContentURIAsBase64(pp.AudioURL.URL)
-					if err != nil {
-						log.Error().Msgf("Failed encoding image: %s", err)
-						continue CONTENT
-					}
-					input.Messages[i].StringAudios = append(input.Messages[i].StringAudios, base64) // TODO: make sure that we only return base64 stuff
-					audioIndex++
-					nrOfAudiosInMessage++
-				case "image_url", "image":
-					// Decode content as base64 either if it's an URL or base64 text
-					base64, err := utils.GetContentURIAsBase64(pp.ImageURL.URL)
-					if err != nil {
-						log.Error().Msgf("Failed encoding image: %s", err)
-						continue CONTENT
-					}
-
-					input.Messages[i].StringImages = append(input.Messages[i].StringImages, base64) // TODO: make sure that we only return base64 stuff
-
-					imgIndex++
-					nrOfImgsInMessage++
-				}
-			}
-
-			input.Messages[i].StringContent, _ = templates.TemplateMultiModal(config.TemplateConfig.Multimodal, templates.MultiModalOptions{
-				TotalImages:     imgIndex,
-				TotalVideos:     vidIndex,
-				TotalAudios:     audioIndex,
-				ImagesInMessage: nrOfImgsInMessage,
-				VideosInMessage: nrOfVideosInMessage,
-				AudiosInMessage: nrOfAudiosInMessage,
-			}, textContent)
-		}
-	}
-
-	if input.RepeatPenalty != 0 {
-		config.RepeatPenalty = input.RepeatPenalty
-	}
-
-	if input.FrequencyPenalty != 0 {
-		config.FrequencyPenalty = input.FrequencyPenalty
-	}
-
-	if input.PresencePenalty != 0 {
-		config.PresencePenalty = input.PresencePenalty
-	}
-
-	if input.Keep != 0 {
-		config.Keep = input.Keep
-	}
-
-	if input.Batch != 0 {
-		config.Batch = input.Batch
-	}
-
-	if input.IgnoreEOS {
-		config.IgnoreEOS = input.IgnoreEOS
-	}
-
-	if input.Seed != nil {
-		config.Seed = input.Seed
-	}
-
-	if input.TypicalP != nil {
-		config.TypicalP = input.TypicalP
-	}
-
-	switch inputs := input.Input.(type) {
-	case string:
-		if inputs != "" {
-			config.InputStrings = append(config.InputStrings, inputs)
-		}
-	case []interface{}:
-		for _, pp := range inputs {
-			switch i := pp.(type) {
-			case string:
-				config.InputStrings = append(config.InputStrings, i)
-			case []interface{}:
-				tokens := []int{}
-				for _, ii := range i {
-					tokens = append(tokens, int(ii.(float64)))
-				}
-				config.InputToken = append(config.InputToken, tokens)
-			}
-		}
-	}
-
-	// Can be either a string or an object
-	switch fnc := input.FunctionCall.(type) {
-	case string:
-		if fnc != "" {
-			config.SetFunctionCallString(fnc)
-		}
-	case map[string]interface{}:
-		var name string
-		n, exists := fnc["name"]
-		if exists {
-			nn, e := n.(string)
-			if e {
-				name = nn
-			}
-		}
-		config.SetFunctionCallNameString(name)
-	}
-
-	switch p := input.Prompt.(type) {
-	case string:
-		config.PromptStrings = append(config.PromptStrings, p)
-	case []interface{}:
-		for _, pp := range p {
-			if s, ok := pp.(string); ok {
-				config.PromptStrings = append(config.PromptStrings, s)
-			}
-		}
-	}
-
-	// If a quality was defined as number, convert it to step
-	if input.Quality != "" {
-		q, err := strconv.Atoi(input.Quality)
-		if err == nil {
-			config.Step = q
-		}
-	}
-}
-
-func mergeRequestWithConfig(modelFile string, input *schema.OpenAIRequest, cm *config.BackendConfigLoader, loader *model.ModelLoader, debug bool, threads, ctx int, f16 bool) (*config.BackendConfig, *schema.OpenAIRequest, error) {
-	cfg, err := cm.LoadBackendConfigFileByName(modelFile, loader.ModelPath,
-		config.LoadOptionDebug(debug),
-		config.LoadOptionThreads(threads),
-		config.LoadOptionContextSize(ctx),
-		config.LoadOptionF16(f16),
-	)
-
-	// Set the parameters for the language model prediction
-	updateRequestConfig(cfg, input)
-
-	if !cfg.Validate() {
-		return nil, nil, fmt.Errorf("failed to validate config")
-	}
-
-	return cfg, input, err
-}

core/http/endpoints/openai/transcription.go

@@ -1,7 +1,6 @@
 package openai
 
 import (
-	"fmt"
 	"io"
 	"net/http"
 	"os"
@@ -10,6 +9,8 @@ import (
 
 	"github.com/mudler/LocalAI/core/backend"
 	"github.com/mudler/LocalAI/core/config"
+	"github.com/mudler/LocalAI/core/http/middleware"
+	"github.com/mudler/LocalAI/core/schema"
 	model "github.com/mudler/LocalAI/pkg/model"
 
 	"github.com/gofiber/fiber/v2"
@@ -25,15 +26,16 @@ import (
 // @Router /v1/audio/transcriptions [post]
 func TranscriptEndpoint(cl *config.BackendConfigLoader, ml *model.ModelLoader, appConfig *config.ApplicationConfig) func(c *fiber.Ctx) error {
 	return func(c *fiber.Ctx) error {
-		m, input, err := readRequest(c, cl, ml, appConfig, false)
-		if err != nil {
-			return fmt.Errorf("failed reading parameters from request:%w", err)
+		input, ok := c.Locals(middleware.CONTEXT_LOCALS_KEY_LOCALAI_REQUEST).(*schema.OpenAIRequest)
+		if !ok || input.Model == "" {
+			return fiber.ErrBadRequest
 		}
 
-		config, input, err := mergeRequestWithConfig(m, input, cl, ml, appConfig.Debug, appConfig.Threads, appConfig.ContextSize, appConfig.F16)
-		if err != nil {
-			return fmt.Errorf("failed reading parameters from request: %w", err)
+		config, ok := c.Locals(middleware.CONTEXT_LOCALS_KEY_MODEL_CONFIG).(*config.BackendConfig)
+		if !ok || config == nil {
+			return fiber.ErrBadRequest
 		}
+
 		// retrieve the file data from the request
 		file, err := c.FormFile("file")
 		if err != nil {